Subscribe to a channel. When it posts a new video, TranscriptHook sends one POST to your n8n workflow with the video details, the full transcript text and timestamped segments. No YouTube API key, no OAuth, no polling code.
Channels are checked about every 10 minutes. Free plan, no card.
The YouTube node gives you metadata: title, description, view count. It does not give you the spoken text of a video. To get that, you would normally set up a Google Cloud project, handle OAuth and quota, write polling logic, and find a way to pull captions.
TranscriptHook does that part. It reads each channel's public RSS feed, waits for YouTube to publish captions, and posts a JSON payload to the Webhook node you already have in n8n. There is no native TranscriptHook node. You use the standard Webhook node.
transcript.available set to false. A video that never gets captions cannot match a keyword filter, so it is not pushed.X-VidProxy-Signature header you can verify with a Code node (see Verifying webhook signatures). Verifying is optional; still keep the n8n path unguessable.No code. Point the two services at each other.
In your workflow, add a Webhook trigger node. Set the HTTP method to POST and copy the Production URL. It looks like https://your-n8n.domain/webhook/[id]. The Test URL only works while you are listening for a test event in the editor.
In the dashboard, click Add channel. Paste the channel URL or @handle, paste the n8n URL as the webhook, and give it a label (it shows up as subscription.label). Then open the subscription's Edit modal and click Send test. It posts a sample payload to your Webhook node so you can map fields before any real video arrives.
Activate the workflow. Use {{ $json.body.transcript.text }}, {{ $json.body.video.title }} and the other fields in any node after the trigger. The Webhook node puts the JSON under body, so the fields live at $json.body.transcript.text and so on.
TranscriptHook reads the channel's public RSS feed, so there is no YouTube API key, OAuth setup or quota to manage. A new subscription starts from the channel's latest video.
The fields in the webhook payload, as n8n expression paths.
// n8n expression paths for common fields // Video info {{ $json.body.video.title }} // "Weekly market wrap" {{ $json.body.video.url }} // "https://www.youtube.com/watch?v=e2evid2" {{ $json.body.video.published }} // "2026-10-02T18:25:51.939Z" {{ $json.body.video.description }} // Full YouTube description, often with links (null on older videos) // Channel info {{ $json.body.subscription.channel_name }} // "Market Brief Weekly" {{ $json.body.subscription.label }} // Your subscription label // Transcript {{ $json.body.transcript.available }} // true / false {{ $json.body.transcript.text }} // Full plain-text transcript {{ $json.body.transcript.segments }} // Array of {start, duration, text} // Only when you set keyword filters {{ $json.body.matched_keywords }} // e.g. ["tariffs"]
Add an IF node on {{ $json.body.transcript.available }} right after the trigger, so videos without captions go to a log step and not to your LLM node.
You assemble these in n8n from the payload. TranscriptHook only supplies the trigger data.
Webhook node → IF node (transcript.available is true) → LLM node (write a five-line summary from transcript.text) → Slack node (post the summary with video.url).
Set keyword filters on the subscription, for example a competitor name or "pricing". The workflow only runs for videos whose transcript matched. Write matched_keywords, video.title, video.url and the first matching segment to Airtable, Notion or Google Sheets. Keyword filters and alerts are on every plan.
Webhook node → Notion node (create a database page). Map video.title to Name, video.url to a URL property and transcript.text to the page content.
Webhook node → OpenAI node (pick a category from transcript.text) → Airtable node (create a record with the video metadata and the category).
Webhook node → OpenAI node (summarize or draft a post from transcript.text) → Gmail or SMTP node (send to your team or list).
Webhook node → Code node (chunk transcript.segments, keeping start so you can link back to the moment in the video) → embeddings node → vector store node. Your retrieval workflow can then answer with a timestamped source.
Free: $0, no card. 1 channel, 1 webhook, 10 on-demand lookups.
Starter: $9 a month. 15 channels, 3 webhooks, 500 lookups per month.
Keyword filters, keyword alerts and the MCP server are on every plan. Monitored channels never use lookups. A lookup credit is used only when an on-demand fetch returns a transcript. Some competitors charge a credit per monitor check. TranscriptHook has no per-check credits, only on-demand lookups. Pro and Agency are not available yet.
{ event: "test", test: true, ... } with placeholder video and subscription data) to your webhook URL, even before you save the subscription. It uses the same delivery path as a real webhook but makes a single attempt with no retry, so you can check your n8n Webhook node before a real video arrives.
{{ $json.body.transcript.available }} equals true. Send the true path to your processing nodes and the false path to a log step. If a video has no captions, TranscriptHook holds it and retries for up to 24 hours, then sends the webhook with transcript.available set to false, so your workflow still sees the event.
subscription.label or subscription.channel_name. Free has 1 webhook endpoint and Starter has 3, so on those plans the Switch node is the usual route.
transcript.available set to false.
subscription.label.
Free includes 1 channel, 1 webhook and 10 lookups. Starter is $9 a month for 15 channels.
No credit card · Free plan